Seems good, thanks!
I'm not familiar with pf's code, but the REASON_SET(&reason, PFRES_MEMORY) was not missing from this patch here?